主講人： Prof. Tzu-Hsiung Yang (Department of Chemistry, National Tsing Hua University)
題目：Deep Generative Models: A New Method for Molecular Discovery
Deep generative models have emerged as a promising way for molecular discovery. They have shown great potential in designing molecules such as organic photovoltaics and drug-like molecules. In this talk, I will first go through a review on the developments of machine learning and deep learning for chemical applications, with the emphasis on their involvement in molecular generation via deep generative models (DGMs). I will then review state-of-the-art DGMs such as variational autoencoder (VAE), generative adversarial network (GAN), recurrent neural network (RNN), and flow. Lastly, I will go through our recent work on applying VAE models to discover optoelectronic materials.